Smith S, Waterall J, Burden ACF. An evaluation of the performance of the NHS Health Check programme in identifying people at high risk of developing type 2 diabetes. BMJ Open 2013;3:e002219. doi:10.1136/bmjopen-2012-002219
There is an error in table 3 of this paper. The error occurred because in calculating specificity in the analysis, the author used DBP cut off of 80 mm Hg instead of 90 mm Hg.
The specificity calculation should therefore be:
Specificity
| Ethnicity | Test-negative & disease-negative patients (d) | Total disease- negative patients (e) | Specificity (d/e×100) |
|---|---|---|---|
| Asian | 2545 | 5223 | 48.7% (47.4% to 50.1%) |
| Other | 3280 | 6297 | 52.1% (50.1% to 53.3%) |
| All | 5825 | 11520 | 50.1% (49.6% to 51.5%) |
All other data analyses and the rest of the data presented are correct.
In addition, there are further corrections as follows:
Abstract (page 1)
Results:
Specificity was 50.1% (95% CI 49.6% to 51.5%)
Results (page 4) Column 2
The NHS Health Check has a specificity of approximately 50% meaning that in the Heart of Birmingham population, 50% of people that were not at risk for diabetes would have been identified by the filter as requiring a blood glucose test.
Discussion (page 6) Column 2
Conversely, half of those that were identified by the filter as being at high risk had HbA1c <42 mmol/mol.